Disruption of Collagen Matrix Alignment in Osteolytic Bone Metastasis Induced by Breast Cancer

Abstract

Breast cancer is highly metastatic to bone tissue and causes osteolytic lesions through osteoclast activation. Although the effects of osteolytic metastasis on bone quantity have been well studied, whether osteoclast activation induced by cancer bone metastasis affects the bone microstructure, a notable aspect of the bone quality, remains uncertain. The aim of this study was to clarify the effect of osteolytic bone metastasis in breast cancer on the microstructure of the bone matrix, particularly the integrity of collagen fibril orientation. Osteolytic breast cancer cells induced hyperactivation of osteoclasts both in vivo and in vitro. Osteoclasts differentiated by culture of monocytes in the cancer cell-derived conditioned medium had an increased number of nuclei; more specific podosome structures were organized compared to osteoclasts differentiated in the control medium. These observations suggest that the resorptive capacity of a single osteoclast was abnormally upregulated in the cancer-involving environment, causing geometrically irregular resorption cavities. Histological studies on mouse femurs with metastasis of breast cancer MDA-MB-231 cells revealed that the osteoclasts in the metastatic bone were abnormally large and they generated resorption cavities that are irregular both in size and in shape. Notably, collagen matrix in newly formed bone in metastatic bone exhibited a significantly disorganized architecture. To the best of our knowledge, this is the first report demonstrating that osteolytic bone metastasis induces the disruption of bone matrix alignment, which determines the mechanical function of bone in both intact and diseased bone tissue.

1. Introduction

Breast cancer is highly metastatic to bone tissue and causes osteolytic lesions that lead to fragility fractures.^1,2) In osteolysis caused by metastatic cancers including breast cancer and multiple myeloma, the major source of osteolytic alterations in breast cancer bone metastasis has been proposed to be the activation of osteoclasts,^3–5) specialized cells that solubilize the bone matrix.⁶⁾ An increased activity of osteoclastic bone resorption affects bone mechanical properties, not only by decreasing the bone mass but also by changing the bone structure.^7,8) Indeed, the molecular crosstalk between osteoblasts and osteoclasts plays a key role in regulating bone metabolism. Although some previous studies have demonstrated the effects of osteoclast activation induced by breast cancer metastasis on the amount of bone,^3,9,10) how it affects the bone microstructure remains uncertain.

It is widely accepted that bone is the composite of collagen fibrils and biological apatite (BAp), and its architecture is highly responsible for the mechanical function of bone.¹¹⁾ Specifically in the long bone, both the collagen fibrils and BAp crystals align in the direction of the bone shaft axis, making the bone highly resistant to loading in the same direction.^12–14) Importantly, this characteristic structure of bone is regulated by the coupling of osteoblasts and osteoclasts.

In a basic bone remodeling cycle, osteoclasts remove old bone and create resorption cavities that are subsequently refilled with new bone matrix by the action of osteoblasts.¹⁵⁾ Indeed, in some skeletal disorders that are accompanied by abnormalities in osteoclasts, the bone microstructure is susceptible to alteration.^16,17) In particular, irregularity of the geometry of osteoclast resorption cavities has been demonstrated in Paget's disease¹⁷⁾ and glucocorticoid-induced osteoporosis.¹⁹⁾ Since these skeletal disorders display an increased resorption activity which is attributable to alterations in osteoclast biology,^20,21) insights into the association of osteoclast biology, geometry of osteoclastic resorption cavities, and microstructure of newly formed bone should be helpful to understand the mechanisms responsible for the maintenance of bone quality in biological systems.

In this study, we focused on the integrity of collagen fibril alignment in metastatic bone as a remarkable aspect of the bone microstructure.²²⁾ Using a mouse model of bone metastasis in breast cancer, we analyzed the influence of breast cancer on osteoclast biology. In vitro osteoclast differentiation was carried out by culturing mouse monocytes either in the cancer-derived conditioned medium (CM) or in the control medium. Differentiated osteoclasts were assessed in the cellular organization that has been proposed to reflect the resorptive capacity of osteoclasts by means of immunofluorescent staining, especially based on the number of nuclei and cytoskeletal organizations including podosome structures. The osteoclastic bone resorption and collagen fibril organization of the metastatic bone in vivo were histomorphologically investigated using birefringence measurement.

2. Materials and Methods

A schematic illustration of the experiments is shown in Fig. 1.

Fig. 1

Schematic illustration of the analysis of the alterations in osteoclast biology and bone microstructure involved in cancer.

2.1 Cell Culture

Human breast cancer cell line MDA-MB-231 (MM-231) was purchased from American Type Cell Culture Collection (Manassas, VA) and maintained in Dulbecco's modified Eagle's medium (DMEM, Nacalai Tesque, Kyoto, Japan) supplemented with 10% fetal bovine serum (FBS, Gibco-Invitrogen, Carlsbad, CA) and 1% penicillin/streptomycin (P/S, Invitrogen), at 37℃ in 5% CO₂.

2.2 In vivo induction of bone metastasis

Five-week-old female BALB/c-nu/nu nude mice were obtained from Japan SLC and housed in a facility with constant humidity and temperature and a 12-h light-dark cycle. They had ad libitum access to standard mouse feed and water. All animal experimentation protocols were approved by the Animal Care and Use Committee of Osaka University.

MM-231 was harvested from 70–80% confluent culture plates and resuspended in phosphate buffered saline (PBS) at a concentration of 10⁶/ml. Then, 0.1 ml of the cell suspension was injected into the left cardiac ventricle of nude mice²³⁾ using 30G needles under inhalation anesthesia with 1.8–2.3% isoflurane (Maylan, Tokyo, Japan). Mice were then monitored three times a week and they were sacrificed 8 weeks after cancer inoculation by administration of an overdose of sodium pentobarbital.

2.3 Isolation of tumor cells from osteolytic bone lesions

Tumor cells from osteolytic lesions were isolated from the hind limbs of cancer-bearing mice as previously described.²⁴⁾ Briefly, tumor cells as well as bone marrow cells were flushed from the resected bones with DMEM and cultured in regular MM-231 culture medium. By continuously cultivating and eliminating the floating marrow cells from the mixture of cells, a pure population of attached cancer cells was obtained two weeks after the cell extraction. This subline of MM-231 was named MM-231-nu and was used in the following in vitro experiments.

2.4 CM preparation

MM-231-nu cells grown to 70–80% confluence were placed in serum-free α-minimum essential medium (α-MEM, Invitrogen) supplemented with P/S for 24 hours. The supernatant was collected, centrifuged at 3000 rpm for 10 min, aliquoted, and frozen at −20℃ until use. Just before use, they were thawed and diluted with the same volume of normal α-MEM and supplemented with 10% FBS and P/S to prepare 50% CM.

2.5 In vitro osteoclast differentiation

Osteoclasts were differentiated from mouse monocytes with the established protocol.²⁵⁾ Bone marrow cells were collected from femurs and tibiae of 8-week-old male C57BL6 mice (Japan SLC) by flushing marrow with α-MEM supplemented with 10% FBS and P/S. They were incubated with red blood cell lysis buffer (Sigma-Aldrich, St. Louis, MO) for 10 min on ice and washed three times with PBS. The obtained bone marrow cells were plated at 3 × 10⁵ cells/cm² and incubated overnight with culture medium supplemented with recombinant human macrophage colony stimulating factor (M-CSF, 25 ng/ml, Peprotech, Rocky Hill, NJ). Nonadherent cells were collected and resuspended either in the normal culture medium or in 50% CM, both of which were supplemented with M-CSF (50 ng/ml) and recombinant mouse receptor activator of nuclear factor kappa-B ligand (RANKL, 50 ng/ml, R&D Systems, Minneapolis, MN). Cells were seeded on conventional plastic substrates placed in a 24-well plate for quantitative analysis of the osteoclast density and the number of nuclei per osteoclast (five wells in each group). For the qualitative assessment of podosome structures, substrates coated with atelocollagen (Nitta Gelatin, Osaka, Japan) that mimic the organic phase of bone were used, since osteoclasts can develop podosome structures only on the bone or the bone-like substrate.²⁶⁾ The medium was changed every two days until day 14 from the cell extraction. The osteoclastic differentiation was confirmed by tartrate-resistant acid phosphatase (TRAP) staining (Wako, Osaka, Japan).

2.6 Visualization of the cellular organization of osteoclasts

The cells were fixed with 4% paraformaldehyde for 20 min, permeabilized with 0.05% Triton in PBS for 15 min. Then, the cells were incubated with Hoechst (Invitrogen) for 90 min and Alexa Fluoro-488 conjugated phalloidin (Invitrogen) for 20 min to stain nuclei and F-actin, respectively. For the assessment of podosome structures, the cells were stained for vinculin prior to the staining for nuclei and F-actin; they were blocked with normal goat serum (Invitrogen) and incubated with the mouse anti-vinculin antibody (Sigma-Aldrich) overnight, followed by incubation with Alexa Fluoro-546 conjugated anti-mouse IgG secondary antibody (Invitrogen) for 90 min. Fluorescent images were acquired using a confocal laser microscopy (Olympus, Tokyo, Japan) and processed using Photoshop 6.0 (Adobe). Cells containing more than three nuclei were identified as osteoclasts. The osteoclast density and the number of nuclei per osteoclast were analyzed in five randomly selected fields at 200× magnification, and the mean values were recorded for each well. P values < 0.05 were considered statistically significant.

2.7 Histological assessment

A tumor-bearing mouse was perfused with 4% paraformaldehyde through the cardiac left ventricle, and femurs were resected and immersed in 4% paraformaldehyde at 4℃ for 7 days. After decalcification with 0.5 M EDTA-2Na solution (pH 7.4) for 7 days at 4℃, specimens were dehydrated through a graded series of ethanol, embedded into paraffin, and then transversely cut into 4 µm-thick sections. Deparaffinized sections were stained with HE or TRAP. TRAP staining was conducted using TRAP kit (Wako) according to the manufacturer's instructions.

2.8 Assessment of collagen fibril orientation

To evaluate the degree of orientation of collagen, a birefringence measurement system WPA-micro (Photonic Lattice, Miyagi, Japan) attached to an upright microscope (Olympus) was used. Deparaffinized sections were imaged with a 20× objective lens. Data were acquired as the average of fifty images, with three settings of circularly polarized monochromatic light (laser wavelength; 523, 543 and 575 nm) for each image. The orientation of polarization axis with the greater index of refraction (the slow axis) of specimen was analyzed with WPA-VIEW software (version 2.4.2.9, Photonic Lattice). Owing to the nature of collagen fibrils as a positive birefringent material, we utilized the direction of the slow axis as the index of collagen fibril orientation.

3. Results and Discussion

3.1 Alterations in the cellular organization of osteoclasts differentiated in vitro with cancer involvement

To examine the effect of breast cancer-derived factors on the cellular organization of osteoclasts on a microscopic level, in vitro osteoclast differentiation was carried out by culturing mouse bone marrow cells either in the cancer-derived CM or in the normal medium. Expecting an enhanced potency of osteoclastogenesis, we used tumor cells extracted from the metastatic bone, because a certain population of cells from a metastatic bone lesion had been reported to display a high incidence of osteolysis.²⁴⁾ TRAP staining on the cultured cells revealed that osteoclast differentiation was successfully achieved in both groups (Fig. 2(A)).

Fig. 2

(A) TRAP staining of the osteoclasts cultured in the control medium and the conditioned medium (CM). Bars = 100 µm. (B) Immunofluorescence images of the osteoclasts cultured on plastic slip. Bars = 50 µm. (C) Osteoclast density and number of nuclei per osteoclast. *P < 0.05, **P < 0.01. (D) Immunofluorescence images of the osteoclasts cultured on collagen-coated substrate. Bars = 50 µm.

As is widely accepted, osteoclast differentiation involves two major events,⁶⁾ namely, cell-cell fusion and the preparation of the resorption apparatuses. These events profoundly influence the cellular organization of osteoclasts by cytoskeletal remodeling; the fusion leads to an increase in the number of nuclei per cell and the cell volume, whereas preparation of the resorption apparatuses accompanies the construction of the actin-based podosome structures. More importantly, these structural properties reflect the bone resorptive capacity of osteoclasts.^27,28) Therefore, we assessed the microscopic structure of the differentiated osteoclasts focusing on the number of nuclei and podosome structures.

For the analysis of the cellular density and the number of nuclei of osteoclasts, osteoclasts were differentiated on conventional plastic substrates (Figs. 2(B) and 2(C)). The osteoclast density in the CM group (99.5 ± 37.3/mm²) was significantly higher compared to that in the control group (41.9 ± 9.7/mm²), consistent with the results of the previous study.²⁹⁾ The number of nuclei per cell was also significantly increased in the CM group (14.7 ± 2.4) compared to that in the control group (4.6 ± 0.9). Since a resorptive capacity of osteoclasts has been reported to positively correlate with the number of nuclei,²⁸⁾ osteoclasts with an increased number of nuclei in the CM group should have an enhanced resorptive capacity.

Furthermore, in the assessment of podosome structures, it became apparent that the CM osteoclasts were organizing thick podosome belts that were quite close to the sealing zone (Fig. 2(D), right). Strong staining for F-actin and vinculin suggested that the podosome cores mainly consisting of F-actin and vinculin had been tightly packed and interconnected with each other, as is the case with the sealing zone formation.^30,31) In contrast, very thin podosome belts were observed in the control osteoclasts (Fig. 2(D), left). As a tight seal of the resorbing area is necessary for efficient bone resorption,^27,32) enhanced development of podosome structures in the CM group is considered to contribute to the excessive activity of osteoclastic resorption.

Taken together, it was suggested that the resorptive capacity of single osteoclasts was upregulated by the breast cancer-derived factors through enhancement of both cell-cell fusion and the development of podosome structures.

3.2 Alterations in the activity of osteoclastic resorption and in the geometry of the resorption cavities in the MM-231 metastatic bone

MM-231 inoculation induced highly osteolytic lesions in the femurs of mice. Cells that are strongly positive for TRAP were detected sequentially throughout the periosteal surface of the MM-231 metastatic bone (Fig. 3(B)), suggesting accelerated differentiation and activation of osteoclasts. HE staining revealed that the osteoclasts in the MM-231 metastatic bone possessed abnormally giant and plump bodies with convoluted membranes that formed the resorption apparatuses. They created resorption cavities that were extraordinarily large and odd-shaped, making the bone surface tremendously ragged (Fig. 3(D), (F)). On the contrary, such polarizing osteoclasts were hardly observed in the diaphysis of the control bone (Fig. 3(A), (C)), and only a few osteoclasts that possess comparatively small and flat bodies with small resorbing areas were detected on the surface of the cancellous bone (Fig. 3(E)). It is likely that the osteoclastic bone resorption was occurring gently in the control bone, keeping the surfaces of bone smooth.

Fig. 3

TRAP staining (A), (B) and HE staining (C)–(F) of longitudinal sections of femurs. Arrowheads indicate osteoclasts. (D) Osteoclasts that possess giant and plumed bodies were occupying the bone surface of the MM-231 metastatic bone. (F) A magnified image of the boxed region in (D). Sturdy protrusions of cellular membrane into the bone matrix (arrows) as well as irregularly large resorption cavities were observed. T: tumor, CtB: cortical bone, CcB: cancellous bone, BM: bone marrow, RC: resorption cavity. Bars = 20 um.

As commonly described, osteoclasts continually remodel their cytoskeleton to organize unique podosome structures that are indispensable for the degradation of both the mineral and organic phases in bone tissue.⁶⁾ Two main podosome structures play roles in the resorption phase; the sealing zone seals the resorption lacunae, and the ruffled border facilitates proton and protease secretion towards the region enclosed by the sealing zone.²⁷⁾ In the osteoclasts of the MM-231 metastatic bone, sturdy protrusions of cellular membrane into the bone matrix were observed (Fig. 3(F), arrows), suggesting that the resorption apparatuses were highly developed due to cancer involvement.

In some osteolytic disorders that are characterized by an enhanced osteoclastic activity, alterations in size and shape of resorptive cavities have been clarified,^18,19,33) suggesting that there is a close relation between the osteoclastic activity and the geometry of the resorption cavities. In our model, it is likely that the hyperactivation of osteoclasts caused the defective regulation in the geometry of resorption cavities in the MM-231 metastatic bone.

3.3 Defective alignment of collagen fibrils in the MM-231 metastatic bone

For the analysis of the organization of the newly formed bone matrix involved in cancer metastasis, the orientation of collagen fibrils near the periosteal surface was evaluated by quantitative birefringence measurement (Fig. 4(A)). Cement lines that appear near the periosteal surfaces were irregularly ragged in the MM-231 metastatic bone (arrowheads, red lines), whereas cement lines were scarcely observed near the periosteum in the control bone (Fig. 4(B), (C)). The matrix between the cement lines and the periosteal surfaces was thought to be the newly formed bone following osteoclastic resorption involved in cancer metastasis.

Fig. 4

(A) Pseudocolor orientation images of longitudinal sections of the cortical bone analyzed by birefringence measurement. Colored bar indicates the orientation of the slow axis from −90 to 90 degrees. Double-headed arrows indicate the orientation of the slow axes that represent the collagen orientation in bone matrix. (B) and (C) Corresponding view of bright-field images of HE (B) and arrow-overlaid images (C). Ragged cement lines (arrowheads, red lines) were observed near the periosteal surface (yellow dotted lines) of the MM-231 metastatic bone. Collagen fibrils were less aligned in the longitudinal direction at newly formed bone (arrows). T: tumor, CtB: cortical bone. Bars = 20 µm. (D) Distribution of the collagen orientation in the fields indicated in (A)–(C).

Although the collagen fibrils in the pre-existing bone mainly aligned in the longitudinal direction in both groups, we found that collagen fibrils were altered to be less aligned in the longitudinal direction at newly formed regions in the MM-231 metastatic bone (Fig. 4(C), right, arrows). Meanwhile, collagen fibrils in the control bone were highly aligned in the longitudinal direction throughout the cortical bone of the diaphysis, keeping the bone surface smooth.

Generally, formation and resorption of bone are coupled both in time and in space.^15,34) After bone resorption is accomplished by osteoclasts, osteoblasts are recruited at the resorption cavities, partially owing to attraction by factors such as demineralized collagen.³⁵⁾ Thus, the traces of osteoclastic resorption form a scaffold for bone formation by osteoblasts. Considering that osteoblasts reside and synthesize the bone matrix dependently on the geometry of the substrate as demonstrated by an in vitro study,³⁶⁾ the manner of osteoclastic resorption may determine the fate of subsequent osteoblastic bone formation.

In this study, the significant disruption of collagen matrix alignment in osteolytic bone metastasis was revealed for the first time. In our model, extraordinarily ragged bone surfaces observed in the MM-231 metastatic bone presumably led to the random configuration of osteoblasts; hence, successively formed bone appeared to be randomly oriented, since we observed the defective alignment of collagen fibrils. In contrast, smoothly resorbed bone surfaces observed in the control bone are likely to result in a smooth monolayer of osteoblasts, and spontaneously, highly aligned collagen fibrils may accumulate around the surface of bone.

4. Conclusion

In conclusion, the current study demonstrated disruption of collagen fibril alignment in newly formed bone in osteolytic breast cancer metastasis. The hyperactivation of osteoclastic resorption, which is attributable to the enhancement of multinucleation and the development of podosome structures in osteoclasts, causes abnormalities in the geometry of resorptive cavities in size and shape in osteolytic bone metastasis. To the best of our knowledge, this is the first report demonstrating the influence of osteolytic metastasis on the microstructure of the bone matrix. Since the microstructure, especially the alignment of the bone constituents, dominantly determines the mechanical function of bone, the present findings of the relationship between osteoclastic activity and bone matrix alignment provide a new insight into the management of osteolytic bone metastasis.

Acknowledgements

This work was supported by JSPS KAKENHI Grant Number JP25220912, 15J01150.

REFERENCES

• 1)  R.E. Coleman: Cancer 80 (1997) 1588–1594. 10.1002/(SICI)1097-0142(19971015)80:8+<1588::AID-CNCR9>3.0.CO;2-G
• 2)  D.L. Waning and T.A. Guise: Clin. Cancer Res. 20 (2014) 3071–3077. 10.1158/1078-0432.CCR-13-1590
• 3)  T. Taube, I. Elomaa, C. Blomqvist, M.N. Beneton and J.A. Kanis: Bone 15 (1994) 161–166. 10.1016/8756-3282(94)90703-X
• 4)  D.R. Clohisy, D. Palkert, M.L. Ramnaraine, I. Pekurovsky and M.J. Oursler: J. Orthop. Res. 14 (1996) 396–402. 10.1002/jor.1100140309
• 5)  N. Giuliani, S. Colla and V. Rizzoli: Exp. Hematol. 32 (2004) 685–691. 10.1016/j.exphem.2004.03.015
• 6)  H.K. Vaananen, H. Zhao, M. Mulari and J.M. Halleen: J. Cell Sci. 113 (2000) 377–381.
• 7)  E. Seeman and P.D. Delmas: N. Engl. J. Med. 354 (2006) 2250–2261. 10.1056/NEJMra053077
• 8)  T. Lin, W. Tong, A. Chandra, S.Y. Hsu, H. Jia, J. Zhu, W.J. Tseng, M.A. Levine, Y. Zhang, S.G. Yan, X.S. Liu, D. Sun, W. Young and L. Qin: Bone Res. 3 (2015) 15028. 10.1038/boneres.2015.28
• 9)  N.C. Hunt, Y. Fujikawa, A. Sabokbar, I. Itonaga, A. Harris and N.A. Athanasou: Br. J. Cancer 85 (2001) 78–84. 10.1054/bjoc.2001.1856
• 10)  J.E. Brown, H. Neville-Webbe and R.E. Coleman: Endocr. Relat. Cancer 11 (2004) 207–224. 10.1677/erc.0.0110207
• 11)  A.K. Nair, A. Gautieri, S.W. Chang and M.J. Buehler: Nat. Commun. 4 (2013) 1724. 10.1038/ncomms2720
• 12)  T. Nakano, K. Kaibara, Y. Tabata, N. Nagata, S. Enomoto, E. Marukawa and Y. Umakoshi: Bone 31 (2002) 479–487. 10.1016/S8756-3282(02)00850-5
• 13)  T. Nakano, K. Kaibara, T. Ishimoto, Y. Tabata and Y. Umakoshi: Bone 51 (2012) 741–747. 10.1016/j.bone.2012.07.003
• 14)  T. Ishimoto, T. Nakano, Y. Umakoshi, M. Yamamoto and Y. Tabata: J. Bone Miner. Res. 28 (2013) 1170–1179. 10.1002/jbmr.1825
• 15)  A.M. Parfitt: Bone 14 (1993) 435–441. 10.1016/8756-3282(93)90176-B
• 16)  J.W. Lee, T. Nakano, S. Toyosawa, Y. Tabata and Y. Umakoshi: Mater. Trans. 48 (2007) 337–342. 10.2320/matertrans.48.337
• 17)  P. Chavassieux, E. Seeman and P.D. Delmas: Endocr. Rev. 28 (2007) 151–164. 10.1210/er.2006-0029
• 18)  D. Chappard, C. Alexandre, J.C. Laborier, J.M. Robert and G. Riffat: J. Submicrosc. Cytol. 16 (1984) 341–8.
• 19)  J. Vanderoost, K. Soe, D.M. Merrild, J.M. Delaisse and G.H. van Lenthe: Calcif. Tissue Int. 92 (2013) 240–250. 10.1007/s00223-012-9674-6
• 20)  S.V. Reddy, C. Menaa, F.R. Singer, A. Demulder and G.D. Roodman: J. Bone Miner. Res. 14 Suppl 2 (1999) 3–8. 10.1002/jbmr.5650140203
• 21)  A. Kukita, C. Chenu, L.M. McManus, G.R. Mundy and G.D. Roodman: J. Clin. Invest. 85 (1990) 1280–1286. 10.1172/JCI114565
• 22)  S. Viguet-Carrin, P. Garnero and P.D. Delmas: Osteoporos. Int. 17 (2006) 319–336. 10.1007/s00198-005-2035-9
• 23)  F. Arguello, R.B. Baggs and C.N. Frantz: Cancer Res. 48 (1988) 6876–6881.
• 24)  Y. Kang, P.M. Siegel, W. Shu, M. Drobnjak, S.M. Kakonen, C. Cordon-Cardo, T.A. Guise and J. Massague: Cancer Cell 3 (2003) 537–549. 10.1016/S1535-6108(03)00132-6
• 25)  I. Boraschi-Diaz and S.V. Komarova: Cytotechnology 68 (2016) 105–114. 10.1007/s10616-014-9759-3
• 26)  S.A. Arkett, S.J. Dixon and S.M. Sims: J. Physiol. 458 (1992) 633–653. 10.1113/jphysiol.1992.sp019438
• 27)  D. Georgess, I. Machuca-Gayet, A. Blangy and P. Jurdic: Cell Adhes. Migr. 8 (2014) 191–204. 10.4161/cam.27840
• 28)  K. Piper, A. Boyde and S.J. Jones: Anat. Embryol. (Berl.) 186 (1992) 291–299. 10.1007/BF00185977
• 29)  K. Tiedemann, O. Hussein, G. Sadvakassova, Y. Guo, P.M. Siegel and S.V. Komarova: J. Biol. Chem. 284 (2009) 33662–33670. 10.1074/jbc.M109.010785
• 30)  P. Jurdic, F. Saltel, A. Chabadel and O. Destaing: Eur. J. Cell Biol. 85 (2006) 195–202. 10.1016/j.ejcb.2005.09.008
• 31)  C. Luxenburg, D. Geblinger, E. Klein, K. Anderson, D. Hanein, B. Geiger and L. Addadi: PLoS ONE 2 (2007) e179. 10.1371/journal.pone.0000179
• 32)  C. Itzstein, F.P. Coxon and M.J. Rogers: Small GTPases 2 (2011) 117–130. 10.4161/sgtp.2.3.16453
• 33)  K. Soe and J.M. Delaisse: J. Bone Miner. Res. 25 (2010) 2184–2192. 10.1002/jbmr.113
• 34)  J. Vanderoost and G.H. van Lenthe: World J. Radiol. 6 (2014) 643–656. 10.4329/wjr.v6.i9.643
• 35)  M.E. Abdelgawad, K. Soe, T.L. Andersen, D.M. Merrild, P. Christiansen, P. Kjaersgaard-Andersen and J.M. Delaisse: Bone 67 (2014) 181–188. 10.1016/j.bone.2014.07.012
• 36)  A. Matsugaki, Y. Isobe, T. Saku and T. Nakano: J. Biomed. Mater. Res. A 103 (2015) 489–499. 10.1002/jbm.a.35189